A copy-paste prompt that edits the slop out of any document
Tim Metz has a command he runs on nearly every AI-generated document that crosses his desk. He calls it “unslop,” and at Animalz, the AI-forward content marketing agency where he serves as director of marketing and innovation, it’s become one of his most-used tools.
The unslop command’s name is a riff on “slop,” Merriam-Webster’s 2025 word of the year meaning “digital content of low quality that is produced usually in quantity by means of artificial intelligence.”
Metz has identified two flavors worth knowing.
- There’s “AI slop” or “content slop”: low-quality AI-generated content produced for volume over value.
- And then there’s “workslop”: internal documents like emails and meeting notes that “seem valuable when you first read them, but then if you really try to understand what they say, you realize they have no value or just don’t make any sense,” as Metz puts it. We’ve all received that email from a colleague that reads as eloquent until you realize it’s needlessly verbose and says nothing.
His “unslop” command exists to catch both. Below, we’ll walk through how it works and how you can use it to maintain high-quality content while still getting the most out of AI.
Where AI earns its place at Animalz (and where it doesn’t)
AI is undeniably changing how marketers do their jobs. Rather than shy away from using AI, Animalz embraces it in both client-facing and internal work. However, the agency doesn’t leverage AI with reckless abandon.
The key to using AI strategically is creating the right infrastructure.
“You really have to build almost a different infrastructure, giving the AI models access to different kinds of data, and I think when you start to do that with the latest models, you start to see very interesting and good results that were not possible before,” Metz says.
At Animalz, this infrastructure looks different depending on the service:
- High-end thought leadership and custom research reports: AI plays a minimal role. These require deep strategic thinking and original analysis.
- LinkedIn content program: This is where Metz has built extensive AI infrastructure, touching nearly every stage of production while maintaining human oversight.
- AEO (Answer Engine Optimization) audits: The team uses significant AI assistance for analysis and recommendations.
The three principles behind the unslop command
Even with careful systems in place, Metz has learned that slop appears anywhere AI touches content creation.
Even if you use AI behind the scenes, for example, for creating brand kits, it tends to produce (work)slop. So it’s not just what comes out of the system, but even if you use AI to create things inside the system (e.g., the brand kit elements that feed into the workflows), those brand kit elements also become sloppy.
To combat slop, Metz developed a custom Claude prompt he calls an “unslop” command. It’s built around three principles: MECE, DRY, and Essential.
Principle 1: MECE (Mutually Exclusive, Collectively Exhaustive)
MECE, a framework from management consulting, ensures content covers everything it needs to without overlap.
“If you follow that rule, you haven’t left anything out, but you also haven’t made different sections overlap each other,” Metz explains.
In practice, MECE means each section should address a distinct topic, and together, all sections should fully address the topic.
Principle 2: DRY (Don’t Repeat Yourself)
This principle is simple: if you’ve said something once, don’t say it again.
Apply this principle by eliminating anything that just rephrases earlier content.
Principle 3: Make things as simple as possible, but not simpler
This principle prevents over-aggressive cutting. The goal is clarity and conciseness, not minimalism that removes necessary context.
Metz notes that each workflow goes through systematic testing after any changes to ensure it doesn’t cut too much.
“Every time you make changes, you need to run standardized tests to see if things are improving or regressing. So if too much has been cut (from a brand kit, for example), you’ll see the quality of the posts starts to drop,” he says.
One giveaway that the command has gone too far is if the text “starts to read like shorthand.” Or if it “starts to cut out examples because it thinks those are unnecessary, but they’re actually very important (in this context of content creation).”
Metz’s “unslop” command
Before and after: What “unslop” actually does to a document
Let’s see Metz’s “unslop” command work its magic on a product strategy roadmap document.
Unslop command initiated via Claude Code
First, it suggests sections to cut and provides justification for doing so. Next, it recommends sections to relocate while providing reasoning.
Claude Code highlights sections to cut and relocate
Finally, Claude Code rewrites the document sans slop.
Claude Code rewrites the document, following the unslop command, and analyzes content reduction and key principles applied
Here’s how the product strategy roadmap document looks pre-unslop command (on the left) and post-unslop command (on the right). The unslopped document gets right to the problem and presents two decisions in a very easy-to-understand way.
Five principles for teams that want to use AI without scaling slop
Metz’s work offers a blueprint for marketing teams that want to scale content without sacrificing quality:
- Match AI use to content type. Not all content should involve AI equally. Animalz uses it extensively for LinkedIn posts and AEO audits but minimally for high-end thought leadership. Before plugging AI into any workflow, know where it adds value and where it doesn’t.
- Build infrastructure before you scale. A prompt is not a system. Create databases that give AI context, build brand kits that define voice and strategy, and establish clear handoff points between AI and humans before volume increases.
- Apply systematic editing principles. Don’t rely on gut feeling to catch slop. Frameworks like MECE and DRY give writers a repeatable standard for cutting filler from both AI outputs and internal documents. Briefs, strategy memos, and brand kits are just as vulnerable as published content.
- Use feedback loops to get smarter over time. Store outputs, analyze what works, and feed that learning back into your prompts and brand kits. AI infrastructure that isn’t updated will degrade in quality.
- Reserve human judgment for what AI can’t replicate. With strong infrastructure handling the mechanical work, writers and editors can focus on strategy, original insight, and the kind of nuanced storytelling that no prompt can produce.
For marketing teams considering their own AI implementations, Metz’s experience shows what’s possible when you combine thoughtful infrastructure with rigorous quality control. Writers can spend less time on mechanical tasks and more time on the strategic and creative work that AI can’t replicate.
Resources to replicate Tim's workflow